Identification and Analysis of the Key Factors Influencing the Increase of Rail Share in Freight Transportation at Ports

One of the objectives of national freight transportation studies is related to analyzing policies aimed at improving freight transportation performance, especially at ports. For this purpose, one of the key tools is mode choice models. In Iran, internal exchanges between the rail and road modes are in competition. This research focuses on mode choice considering the need to increase the rail share through the establishment and strengthening of the multimodal transportation network and improving the flexibility of freight transportation. Subsequently, with access to key factors influencing the mode choice in freight transportation and the data received from Iran Railways, the Road Maintenance and Transportation Organization, and the Ports and Maritime Organization of Iran, machine learning algorithms were employed to compare the highest accuracy among them. Furthermore, in order to achieve improvement and increase the rail share in freight transportation, significant features were extracted with a focus on increasing the rail share, leading to desirable results. The results indicated that the XGBoost algorithm had the highest accuracy among the algorithms. Using this algorithm and the method for extracting important features in the goods dataset, it was found that the most influential features in increasing the rail share are: road ton-kilometers based on the total maximum road tonnage, rail ton-kilometers based on the total maximum road tonnage, maximum road tonnage, maximum rail tonnage, total rail tariff (transportation and loading/unloading), and the value of goods transported between origin and destination.